Any recommendations for contract assembly companies?

So your expecting to give a company $8000 and receive 1000 units of a product in sale ready unit? I don't see how this is possible. How many thru hole pins/surface mount pads? Custom molded enclosure I assume? How many assembly parts? Blister pack for sales display required? If you work with the manufacturers directly it may be possible, but contractors can burn you money pretty quickly.

I can say with some confidence you're not going to find anyone in America to do it for that price, and you're not going to find anyone in Asia willing to handle such a tiny volume unless this is the pilot run for something much larger. 1000 units is nothing, and it's the hundredth root of nothing if the end product has a mfd cost of $8. By packaging I assume you mean taking the populated PCB, putting it in a housing (included in your $8?) and then retail-packaging the end product?

If you intend to put this into actual commercial production then build a relationship with a Far East CM and do one-stop shopping. If 1000 units is a one-shot with no guarantee of ongoing large orders then accept higher prices.

I assume he means packaging of the stuffed PCBs (for shipment), which can add noticably to the assembly cost, both for setup and for unit cost. He doesn't mention testing.

Best regards, Spehro Pefhany

Oops, I missed the 'sales-ready' bit.

Yeah, he needs to add some zeros here and there. 80K would be more like it, for a relatively simple product without too much tooling.

Best regards, Spehro Pefhany

You want to have someone ....

1. Make a pcb
2. Supply the components
3. Assemble the pcb
4. Test it
5. Supply an enclosure
6. Assemble the stuffed pcb into the enclosure
7. Supply a carton suitable for point of sale ( colour printed ? )
8. Package the product in the carton

And you want this for $8 on a run of only 1000 pcs ? With no tooling or set-up costs ?

You're dreaming ! Unless it's a *tiny* item. You haven't specified what it is.

It is probably best to one-stop shop. You clearly can't afford western prices so you need to go to Asia. I have some trouble figuring why any outfit would be interested in an $8000 dollar job though. You might have some luck if you wanted 10,000 pcs.

Or get it made in the west at a higher price.

Graham

I guess I didn't explain very well. The $8 target cost includes ONLY the PCB, population/reflow, and assembly of the final product. All of the components, the case, and other parts of the product will be supplied by me. This is service-only expense. The exact steps of the process are:

Should be possible. You forgot test fixtures and testing.

Best regards, Spehro Pefhany

Yes, that issue did occur to me. I should be able to make a jig that will test a populated board for functionality in about 1 second. Could probably make a jig to test 10 at a time easily. Mike Poulton

Thanks! I'll contact them. Unfortunately, there are very few viable choices in the design of this product. It consists of two buttons, a

Bass-ackwards. It is normal to test boards while they are still panelized.

It's not a consumer product. A run of 1000 is approximately 4 sales orders worth of product, or about 3 months of projected sales. Marketing is primarily through direct industry contacts. Mike Poulton

My experience is that you already have to have the assembly house chosen before you even draw a schematic. If you want something really cheap, the best is to dovetail onto a product the house is already building. What parts you choose-- leaded, SM, etc. depend greatly on the processes used in manufacturing.

For low volume work (like your 1000 pieces), try D.E.C. in Santa Ana, CA (1790 E. McFadden, Unit 105). They do nice work for smallish runs.

good luck.

If the total quantity is only 1,000, and test time is negligible, I'd test them one at a time. Even $1K in fixture is $1/unit.

Best regards, Spehro Pefhany

How do you plan on marketing such a small run- the Internet?

Well, it depends if it's ICT or FT. FT I guess is OK one by one. But it costs little more to build an ICT jig for a panel of 50 vs a single circuit, since you can use 50 sets of probes switches by relays to the tester.

With such a low production rate you should consider employing someone to make them. 1 part time employee should be able to keep up with that demand. You need to have the board itself made externaly.

Qualify "test"... ICT's purpose is to make sure all desired locations are populated with approximately the correct component values. Something like a Zentel tester is used. The bed-o-nails is custom but the tester itself is generic, software-reprogrammable. It's not significantly more expensive to make a bed-o-nails for the panel than for the individual PCB; it's a step and repeat CNC process to drill the probe positions.

Sometimes ICT will include programming flash chips, EEPROMs, serialization, etc.

FT's purpose is to make sure the device as a whole functions within specifications. This is almost always a custom fixture.

It's very difficult to perform equivalent functionality to a bed-o-nails ICT in vivo on the device itself. If this feature is a customer requirement, then it can be engineered in, but at a staggeringly high cost.

If you do the enigineering work on that test set for free, yes. I have designed many test sets, involving HW engineers and SW folks. When it was all said and done the cost was north of $30k at western rates. It's not just R&D and build, it needs to be ECOed and that costs.

Best is to built a really good test functionality into the circuit itself. IOW where it can test itself on the first power-up.

Regards, Joerg

Hello,

(sorry, don't know your name)

With test I mean a detailed enough diagnostic routine. In most of our cases it entails an extra few routine on a uC. This pipes signals from otherwise quiet port pins into strategic points and from the response it is deduced what could be wrong. Mis-stuffed caps, resistors, lack of hfe in a transistor and so on. The transistor thingie is next to impossible with ICT, at least for RF. Not that we hadn't done that on a Zehntel in the past but it required elaborate 'satellite boards'.

Wrong component values are pretty much a thing of the past these days. Same with tombstoning but that could be found via optical inspection anyway. The last time I saw a design being subjected to the big Teradyne was more than a decade ago.

Yes, that's a nice advantage. But contract assemblers charge a lot for ICT.

There usually is no fixture. Upon power-up plus sometimes a 'secret button' press the circuit performs a self-test. The only time when there is a fixture is for checking cal levels of voltages, frequencies and the like. But that boils down to a connection plus connector saver and some PC controlled lab equipment.

True but one doesn't necessarily need all that functionality. In the RF world the functionality of ICT is actually lower or sometimes ICT can't do it at all.

Staggering cost? Cannot agree here. Not if testability is a functional requirement from day one of the design cycle. Which it should. In the medical world that's almost the law so we have to.

Regards, Joerg